Compare band combinations
First, you'll take an initial look at the burn scar using Landsat 8 imagery. After acquainting yourself with the study area, you'll change the band combination of the imagery to better see the burned areas. Then, you'll create a custom combination to emphasize burn scars.
Open the project
Before you begin your analysis, you'll download and open a project package containing the raw data for your assignment.
- Go to the
Montana Fires item details page.
Montana Fires is a project package. Project packages contain all maps, data, folders, and toolboxes for a project.
- Click Download. If prompted, download the file to a location you can easily remember.
Next, you'll open the project in ArcGIS Pro.
- Start ArcGIS Pro. If prompted, sign in using your licensed ArcGIS account.
If you don't have ArcGIS Pro or an ArcGIS account, you can sign up for an ArcGIS free trial.
When you open ArcGIS Pro, you're given the option to create a new project or open an existing one. If you've created a project before, you'll see a list of recent projects.
- Click Open another project (if you've used ArcGIS Pro before) or Open an existing project (if you haven't).
Next, you'll search for the project you downloaded.
- In the Open Project window, browse to the location where you downloaded the Montana_Fires project package. Double-click the package to open it.
The project opens with two layers in the Contents pane. The two layers are imagery layers and are turned off. The map is centered around Glacier National Park in Montana, which comprises more than one million acres. Fires are a natural part of the regional ecology, but it is critical for forest management services to track their extent.
Enhance the imagery
Now that you've accessed the project data, you'll look at and enhance the imagery. The project came with two Landsat 8 imagery layers, clipped to a study area around two particular wildfires: the Thompson Fire and the Reynolds Creek Fire. Both images were captured in August of different years.
- In the Contents pane, check the box next to the 2014 layer to turn it on. Right-click the layer and click Zoom To Layer.
The image is dark and difficult to see. You'll adjust the brightness, contrast, and gamma to better see the image. Brightness determines whether an image is lighter or darker. Contrast determines how distinguishable features are from one another. Gamma determines the relationship between how an image detects light and its actual luminescence. Increasing all of these will improve the visibility of the image.
- If necessary, in the Contents pane, click the 2014 layer to select it.
- On the ribbon at the top of the application, click the Appearance tab. In the Enhancement group, increase Layer Brightness to 20, Layer Contrast to 25, and Layer Gamma to 1.8. Press Enter.
The changes occur immediately on the image.
You can now see the environment in more detail. The mountainous terrain is marked by valleys and lakes. Some of the peaks are snowcapped while others are obscured by cloud cover. Since this image was taken in August, the mountains may have glaciers or permanent snow. The terrain also appears to have ample vegetation. The type of vegetation and the slope of the mountains affect fire, particularly the speed at which it spreads. This is what the area looked like in 2014, before the Reynolds Creek and Thompson fires. Next, you'll take a look at the 2015 imagery.
- In the Contents pane, uncheck the 2014 layer to turn it off. Check the 2015 layer to turn it on.
The default appearance of the 2015 image is clearer than the 2014 image was, but it could still use some brightness, contrast, and gamma correction.
- In the Contents pane, click the 2015 layer to select it.
- On the Appearance tab, increase Layer Brightness to 10, Layer Contrast to 15, and Layer Gamma to 1.5.
The 2015 imagery has two distinct differences from the 2014 imagery. First, a large gray cloud covers the central-southern portion of the image. This cloud is actually smoke from the Thompson fire, which was still burning when this image was taken.
Secondly, to the upper left of the lake in the central-northern part of the image is a long reddish streak. This is the burn scar of the Reynolds Creek fire, which had stopped burning by the time this image was taken.
While both fires are visible, their exact boundaries are unclear. Next, you'll change the band combination of the imagery to emphasize the fires.
View different band combinations
Landsat imagery measures ranges of wavelengths of the electromagnetic spectrum, including some that are invisible to the human eye. These ranges are called spectral bands. The bands are described in the following table:
|Number||Name||What this band shows best|
Shallow water, fine dust particles
Deep water, atmosphere
Man-made objects, soil, vegetation
Shortwave Infrared 1
Cloud penetration, soil and vegetation moisture
Shortwave Infrared 2
Improved cloud penetration, soil and vegetation moisture
Black-and-white imagery, crisper detail
Thermal Infrared 1
Thermal mapping, estimated soil moisture
Thermal Infrared 2
Improved thermal mapping, estimated soil moisture
Bands 2, 3, and 4 (Blue, Green, and Red) make up the spectrum of light visible to the human eye. The Natural Color band combination, which your imagery currently uses, combines these three bands to approximate how imagery would look to a person. Next, you'll change the band combination to emphasize the fires and better see their boundaries.
- In the Contents pane, confirm that the 2015 layer is selected.
Under the layer name are the bands the image currently uses: the Blue, Green, and Red bands that encompass visible light. The Red and Green bands emphasize vegetation, which can be useful for seeing fires because of the contrast between highly vegetated areas untouched by fire and areas where vegetation has been destroyed by fire. Using another band that emphasizes vegetation, like Near Infrared (band 5), could improve the contrast.
- On the Appearance tab, in the Rendering group, click Band Combination and choose Color Infrared.
The image changes to show the new band combination. In the Contents pane, the bands beneath the layer name also change, indicating that this image combines Near Infrared, Red, and Green bands (3, 4, and 5).
In this image, vegetation appears as red. Both fire areas appear as dark brown. Compared to the original image, the fires appear more clearly, especially the Reynolds Creek fire north of the lake. However, the Thompson fire is still obscured somewhat by smoke. Next, you'll try a band combination using the Shortwave Infrared bands (6 and 7), which penetrate clouds.
- On the Appearance tab, click Band Combination and choose Land/Water Interface.
The bands beneath the layer in the Contents pane change to both Shortwave Infrared bands and the Near Infrared band, meaning this image combines bands 5, 6, and 7. Although the main purpose of this combination is to delineate land and water, it also penetrates haze (or, in this case, smoke). Almost no smoke appears around the Thompson fire, making its boundaries much clearer. However, the burned areas appear orange while the surrounding mountain slopes appear yellow. This makes the Reynolds Creek fire, which spreads into the mountains, more difficult to see.
- Click Band Combination again and choose Vegetation Analysis.
This combination uses the Red, Near Infrared, and Shortwave Infrared 1 bands (4, 5, 6). It thus combines the emphasis of vegetation from the Color Infrared combination with some of the haze penetration of the Land/Water Interface combination. Although some smoke is visible around the Thompson fire and the Reynolds Creek fire blends somewhat into the mountain slopes, these issues are less severe than in the previous combinations.
If the haze could be reduced just a little more, this image would probably be the best for digitizing the burn scars. However, none of the remaining default band combinations improves on the three you looked at. To tailor the imagery to your needs, you'll create a custom band combination.
Create a custom band combination
So far, you've used preconfigured band combinations. Next, you'll choose your own bands to create a custom band combination that will improve on the Vegetation Analysis band combination by further reducing haze.
- In the Contents pane, locate the 2015 layer.
The Vegetation Analysis band combination uses the Shortwave Infrared 1 band to reduce haze and the Near Infrared and Red bands to emphasize vegetation. Switching Shortwave Infrared 1 to Shortwave Infrared 2 would improve the haze (or cloud) penetration.
Since bands like Near Infrared and Shortwave Infrared cannot be seen, imagery displays them as RGB (Red Green Blue) composites. This means that red light or color is used to display the imagery in the band with which it's associated, and so on for the green and blue bands. Any band can be used in any of the three composite colors, which is why it's possible to have the Red band displayed using blue light. The color of the symbol and the first word indicates the composite color, while the second word indicates the band.
- Right-click the ShortWaveInfrared_1 band and choose ShortWaveInfrared_2.
The Panchromatic band (band 8) and both Thermal Infrared bands (bands 10 and 11) have been removed from the data, which is why they do not appear in the list of bands.
The band automatically changes, as does the image on the map. Since the change was between two similar wavelengths, the differences on the map are not incredibly obvious, but change did occur. Next, you'll replace the Red band with the Blue band.
- Right-click the Blue color and choose Blue.
The image on the map changes slightly. Although still visible, the haze has been reduced.
To easily access this band combination in the future, you'll save it as one of the default combinations on the Appearance tab.
- On the Appearance tab, click Band Combination and choose Custom.
The Custom Band Combination window opens.
- For the Red color, choose ShortWaveInfrared_2. For the Green color, choose NearInfrared. For the Blue color, choose Blue.
- Name the custom band combination Burn Scar Analysis and click Add.
The band combination is added to the drop-down menu when you click the Band Combination button, allowing you to quickly apply it to other imagery (or reapply it to this image if you change the band combination again).
You can remove custom band combinations from the list only after starting a new session of ArcGIS Pro.
- Save the project.
You've displayed imagery of two fires in Glacier National Park, first as a natural color image and then using other band combinations that emphasized the burned areas better. Ultimately, you created a custom band combination specifically meant to highlight burn scars. Although this custom combination is more suitable for deriving exact burn scar boundaries, it still requires some visual interpretation in order to determine what is burned and what is not. Next, you'll use a mathematical formula called a burn index to calculate burned areas quantitatively, providing an even more exact measure of where the fires raged.
Calculate the burn index
Previously, you looked at the imagery through different spectral band combinations to visually identify burn scars. Next, you'll use an equation to quantitatively identify burned areas. This equation is the Normalized Burn Ratio (NBR). It mathematically compares the Near Infrared and Shortwave Infrared 2 bands (bands 5 and 7, respectively) to determine burn severity. Then, you'll compare the NBR of the 2014 and 2015 imagery to calculate NBR change, showing only areas that have been burned between the dates both images were taken. You'll then digitize the burned areas as feature classes and share those feature classes on ArcGIS Online.
Calculate the Normalized Burn Ratio
You'll calculate the NBR twice: once for the 2014 image and once for the 2015 image. To make the calculation, you'll use the Raster Calculator geoprocessing tool and use the following equation:
NBR = (Band 5 – Band 7)/(Band 5 + Band 7)
For the equation to work, you'll extract the bands used in the calculation (bands 5 and 7) from the original data.
- If necessary, open your Montana Fires project.
- On the ribbon, click the View tab and click Catalog Pane.
The Catalog pane opens (it may have been open already). Next, you'll navigate to the folder where the data is stored. Because you downloaded this project as part of a package, a folder connection has already been established.
- Click the drop-down arrow next to Folders. Open the montana_fires folder, the commondata folder, and the raster_data folder.
The raster_data folder contains two raster datasets: the imagery layers currently on the map.
- Click the arrow next to the G_2014.tif file.
The individual bands are listed. You need band 5 (Near Infrared) and band 7 (Shortwave Infrared 2).
- Right-click the NearInfrared band and choose Add To Current Map.
The band is added to the map. It looks like a black-and-white image because it is not being shown as part of an RGB composite. Single-band image layers often look black and white.
- Right-click the ShortWaveInfrared_2 band and choose Add To Current Map.
- Click the arrow next to the G_2015.tif file. Add the NearInfrared band and the ShortWaveInfrared_2 band to the map.
The Contents pane now has four band layers, two for each year. Next, you'll use these bands to calculate the NBR with a geoprocessing tool called Raster Calculator.
- On the ribbon, click the Analysis tab and click Tools.
The Geoprocessing pane opens.
- In the Geoprocessing pane, click the search box and type Raster Calculator. From the list of results, click Raster Calculator (Spatial Analyst Tools) or Raster Calculator (Image Analyst Tools) (either will work).
The Raster Calculator tool opens. This tool allows you to create a new raster dataset based on an equation to determine its pixel values. You can create this equation with existing images in your project, including the individual bands from the fire imagery. You'll run this tool twice, once for the 2014 bands and once for the 2015 bands, using the following equation:
(Band 5 – Band 7)/(Band 5 + Band 7)
- Under Rasters, double-click G_2014.tif_NearInfrared to add it to the expression box.
The 2014 Near Infrared band will be the Band 5 in the equation.
- Under Operators, double-click the minus sign to add it to the expression. Then, double-click the G_2014.tif_ShortWaveInfrared_2 raster.
To create a ratio, you'll divide what you have so far by the same two bands added together instead of subtracted.
- Edit the expression box by adding parentheses around the existing expression. After the closed parenthesis, add a division operator.
- Double-click the G_2014.tif_NearInfrared raster to add it to the expression. Add an addition operator and then double-click the G_2014.tif_ShortWaveInfrared_2 raster.
- Put parentheses around the second part of the equation.
The completed expression reads ("G_2014.tif_NearInfrared" - "G_2014.tif_ShortWaveInfrared_2") / ("G_2014.tif_NearInfrared" + "G_2014.tif_ShortWaveInfrared_2").
To increase the size of the expression box and see the entire expression at once, increase the size of the Geoprocessing pane by dragging its border.
Before you run the tool, you'll set the output location and name.
- For Output raster, click the Browse button.
- In the Output raster window, under Project, click Folders. Double-click the montana_fires folder.
Because Montana Fires is a project package, its folder also includes the commondata and p20 folders. You can ignore these.
- Change the name to 2014_nbr and click Save.
- Click Run.
The tool runs and adds a new image to the map. Like the bands, the new layer is black and white and gives little information about burn scars. Only when you determine how much the NBR changed between 2014 and 2015 will the burn scars become apparent. To make this comparison, you need to use the Raster Calculator tool again for the 2015 bands.
- In the Geoprocessing pane, under Map Algebra expression, change all instances of 2014 to 2015.
- For Output raster, change the output name to 2015_nbr (leave the output location unchanged) and click Run.
Determine change in NBR
Next, you'll use the Raster Calculator tool one more time to calculate the change in NBR between the two images. In doing so, you'll remove values for areas that were not burned between 2014 and 2015, showing only the burn areas and removing everything else from the image.
- In the Geoprocessing pane, delete the expression in the Map Algebra expression box.
You want to find the change, or difference, between the two NBRs. To do so, you'll subtract the post-fire (2015) NBR from the pre-fire (2014) NBR.
- In the Map Algebra expression box, create the expression "2014_nbr" - "2015_nbr".
- Change the output raster to change_nbr (leave the output location unchanged) and click Run.
The fire locations now appear almost solid white, contrasting strongly against the gray and black areas around them. The only other white areas are the snowy mountainous areas, which are generally not conterminous with the fire areas. You can heighten the contrast by symbolizing the image.
- In the Contents pane, under the change_nbr layer, click the color ramp.
The Symbology pane opens. The change_nbr layer symbology is determined by a color ramp instead of an RGB composite because, like the other NBR layers, it contains only one band.
- In the Symbology pane, next to Color scheme, choose the Condition Number (green to red) color scheme.
To see the name of a color scheme, either point to the scheme or check the Show names box at the bottom of the Color scheme menu.
The symbology updates.
Both fires are now clearly demarcated on the map.
- Close the Symbology pane and the Geoprocessing pane.
You no longer need the original NBR layers or the individual band layers, so you'll remove them.
- In the Contents pane, right-click the 2015_nbr layer and choose Remove.
- Remove the following layers:
While pressing Shift, click the first and last layers listed to select all the layers. Right-click one of the selected layers and choose Remove.
Only the NBR change layer, the two original imagery layers, and the Topographic basemap remain.
Digitize the fire area
You now have a clear enough image of the fire extents to digitize them as features to share with the Montana Department of Forestry and Resource Management. You'll first create a feature class and then use editing tools to approximate the boundaries of both fires.
- In the Catalog pane, open the Databases folder.
The folder contains a geodatabase called montana_fires.
- Right-click montana_fires, point to New, and choose Feature Class.
The Geoprocessing pane opens to the Create Feature Class tool.
- In the Create Feature Class pane, for Name, type Fires. Leave the other parameters unchanged and click Finish.
The new feature class is created.
- In the Catalog pane, click the arrow next to montana_fires.gdb to expand it. Right-click the new Fires feature class and choose Add To Current Map.
- In the Contents pane, click the symbol for Fires.
The Symbology pane appears. Because the new feature class's default symbology has a solid fill, it can be difficult to properly draw all around the map feature you're tracing.
- In the Symbology pane, on the Gallery tab, choose the second option, Black Outline (2pts).
- Zoom to the Reynolds Creek fire.
- On the ribbon, click the Edit tab. In the Features group, click Create.
The Create Features pane appears. It contains the layers for which you can create new features.
- In the Create Features pane, click the Fires feature layer.
The pointer changes to a crosshair when you move it over the map.
- Click anywhere on the edge of the Reynolds Creek fire area to begin drawing a polygon feature.
If you click anywhere else, another vertex will be placed.
- Add vertices along the edge of the fire area.
The more vertices you add, the more accurate your feature will be. Since you're only doing an exercise, don't worry about creating a perfect feature, just one that's reasonably accurate.
- When you've finished placing vertices, double-click to finish creating the feature.
If you don't like how your feature turned out, click Edit Vertices (on the Edit tab, in the Tools group) and modify the placement of vertices. Alternatively, on the Edit tab, in the Features group, click Delete to delete the entire feature and start over.
- When you're satisfied with the feature, on the Edit tab, in the Manage Edits group, click Save.
- In the Save Edits window, click Yes to save all edits.
- Press the Esc key to get back to the map navigation mode.
- Zoom out and zoom back in to the Thompson fire.
The Thompson fire is larger than the Reynolds Creek fire, so it will take longer to digitize.
- In the Create Features pane, under Fires, click the Polygon button.
- Digitize the Thompson fire.
- When you're satisfied with your feature, save the edits.
- Close the Create Features and Symbology panes. Return to the full extent of the imagery.
Add attribute information
You've created features for both fires, but they currently have no attribute information. You'll edit the attribute table to identify each fire and calculate each fire's acreage.
- In the Contents pane, right-click the Fires layer and choose Attribute Table.
The attribute table has two features, in the order they were digitized. The last feature you created, the Thompson fire, might still be selected.
- On the attribute table ribbon, click the Clear Selection button.
The perimeter and area of the fires have already been calculated, but these calculations are in square meters. A more standard measurement for area would be acres.
- On the attribute table ribbon, click the Add Field button.
The Fields view opens, with an empty field at the bottom. You'll add two fields: one for the fire name and one for the fire acreage.
- For the new field, change the Field Name to Name. Double-click the Data Type cell and choose Text.
- On the ribbon, in the Changes group, click Save.
If you have unsaved edits from when you digitized the fire features, you won't be able to save changes to the attribute table. If you can't save, close the attribute table without saving, save your feature edits on the Edit tab, and add the new field again.
- Click the bottom of the list of fields to add a new field.
- Change the name of the new field to Acres and the Data Type to Float.
- On the ribbon, click Save.
- Close the Fields view to return to the attribute table.
The fields are currently empty. You'll edit the Name fields directly, but to calculate acreage you'll run a geoprocessing tool.
- Double-click the Name field for the first feature to edit it. Type Reynolds Creek and press Enter.
- Change the name of the second feature to Thompson.
- Right-click the Acres field heading and choose Calculate Field.
The Geoprocessing pane opens to the Calculate Field tool. This tool allows you to create an expression to determine field values. The Fires feature class already has an area field, but it's in square meters, not acres. One acre equals 4,046.86 square meters, so you'll use this conversion rate to calculate acreage.
- Under Expression, double-click Shape_Area to add it to the expression box. Add a division operator and type 4046.86 after it.
- Click Run.
The Acres field is calculated. The Reynolds Creek fire is approximately 4,400 acres, and the Thompson fire is approximately 12,400. Your values will vary because you digitized your features differently.
- Close the attribute table and the Geoprocessing pane.
- On the ribbon, click the Edit tab. In the Manage Edits group, click Save and save the edits you made to the attribute table.
- Save the project.
Share your results
You now have polygon features for both fires with attribute information on their names and acreage. The last thing you'll do is publish the Fires feature class to ArcGIS Online to share it.
- In the Contents pane, right-click the Fires layer, point to Sharing, and choose Share As Web Layer.
The Share As Web Layer pane appears. Before you can share a layer, you must input metadata so it can be searched for and catalogued.
- In the Share Web Layer pane, change the name from Fires to Glacier_National_Park_Fires. Add your name or initials to the end of the name to make it unique.
- For the summary, copy and paste the following text:
Perimeter definition of the Reynolds Creek and Thompson fires in Glacier National Park during the summer of August 2015. Perimeters defined by difference in Normalized Burn Ratio.
- For the tags, type Fire, Reynolds Creek, Thompson, Glacier National Park, Montana and then press Enter.
- For Sharing With, choose to share with either Everyone or your organization, depending on who you want to see your web layer.
- Click Analyze.
The layer is analyzed for errors. If metadata is missing or there is something wrong with the data, the error will be catalogued and described so you can fix it.
- If no errors are found, click Publish.
The layer is published to ArcGIS Online. You can access it in your Content page. The layer can be added to any number of maps, symbolized, and shared.
In this lesson, you used Landsat imagery to determine the extent of two fires. You first looked at the imagery through various spectral band combinations to visually assess the fire location. Then, you calculated the Normalized Burn Ratio to specifically highlight burned areas. Lastly, you digitized both fires and shared them to ArcGIS Online. In a real-world scenario, the Montana Department of Forestry and Resource Management could then use your layer for vegetation succession studies or to plan for future fires in the area.
You can find more lessons in the Learn ArcGIS Lesson Gallery.